Update Jan. 25, 2023:
Connie Guglielmo, editor-in-chief of CNET, releases another statement about AI articles published since November: According to the note, 77 short articles were created using AI, or about one percent of the content published since November. Editors first created drafts using AI, then edited and expanded them before publishing.
Editors reviewed all published AI articles after an error was discovered and criticized externally: they found incorrect information in 41 of the 77 articles. A “small number” of articles required major corrections, and several required minor corrections, such as incomplete company names, transposed numbers, or inaccurate wording, according to Guglielmo.
Use of the AI tool will resume once editors are confident that editorial processes are preventing human and AI errors. CNET also promises more transparency and better plagiarism checks.
“We’re committed to improving the AI engine with feedback and input from our editorial teams so that we – and our readers – can trust the work it contributes to,” Guglielmo wrote.
We also learned there’s a lot of fear, uncertainty, doubt & misinfo around AI engines in publishing, esp since the debut of ChatGPT. But we & many media orgs, publishing companies and content creators, know the trend is toward more, not less, adoption of automated tech like AI.
— Connie Guglielmo (@techledes) January 25, 2023
Original article from Jan. 23, 2023:
CNET uses AI to write generic, sometimes incorrect search engine copy on financial topics. The US tech magazine stops the practice after widespread criticism.
In mid-January, search engine content specialist Gael Breton discovered that CNET had published more than 70 financial advice articles optimized for Google using “automated technology”. The articles were published under the name “CNET Money Staff”, which was later changed to “CNET Money”. This is not a clear attribution to a robo-editor.
CNET confirmed its use of AI technology for general financial stories shortly after Breton’s revelation. The magazine said it wanted to “see if the tech can help our busy staff of reporters and editors with their job to cover topics from a 360-degree perspective.”
Although CNET said the copy was thoroughly reviewed before publication, some of it contained glaring errors. This is especially bad because personal finance articles often include ads for financial products, some on a pay-for-performance basis.
CNET backs away from AI-generated copy
According to a revelation by CNET competitor The Verge, executives at the publisher have now communicated internally that they no longer want to publish AI-generated texts for the time being. Editor-in-chief Connie Guglielmo says the accusation that CNET secretly published the AI texts is inappropriate: The magazine did it “quietly.”
The CNET team is said to have had little insight into the technology used. Red Ventures reportedly developed the system. The publishing group acquired CNET in 2020. The CNET management team plans to provide employees with further information about the AI tool and a preview version in the coming days.
Other SEO portals such as Bankrate and Creditcards.com are also said to be forgoing AI-generated content for the time being. Like CNET, they are also owned by Red Ventures.
Red Ventures also developed a text tool that automatically inserts current numbers, such as interest rates, into financial articles. That tool will remain in operation.
CNET as a role model for SEO spammers
CNET’s AI text strategy, successful at least in Google’s rankings, forced Google’s search team to justify itself, as it had previously announced that it would treat AI-generated text like spam.
Google search spokesman Danny Sullivan explained CNET’s good rankings by saying that Google focuses on “useful content” rather than how a piece of content is created. The goal is to remove useless content from search, Sullivan said. Whether it’s written by humans or machines doesn’t matter, he added.
Some SEO professionals interpret Sullivan’s statement as a free pass to target masses of AI-generated text for good Google rankings. Even the errors in CNET’s text had no effect on Google’s search bot.
In addition to ChatGPT, this poses another risk to Google’s core business: If the search giant does not get a handle on AI spam, it will lead to a devaluation of Google’s search results and thus hit its main source of revenue.
Caught between chatbot search and the threat of AI spam, Google needs to adapt its products to a new era. According to the New York Times, this transformation is already well underway.
Whether CNET’s success is so easy to replicate for SEOs is questionable: The rankings of the AI copy benefit as much as the human copy from CNET’s well-known and highly linked publishing brand.
It’s possible that Google will manage to treat AI text differently depending on where it’s coming from. This would give publishers more leeway in automation techniques, while AI spam could still be limited by Google.
However, this process would likely require an enormous amount of coordination on Google’s part and would hardly be fair. In addition, Google could be forced to justify itself if it were to use AI text tools for search results itself in the future, for example in the form of a chatbot.